BerriAI/litellm · error · OobaboogaError

str(e)

Error message

str(e)

What it means

After the error checks, the oobabooga transformation copies `completion_response['choices'][0]['message']['content']` into the model response. Any shape mismatch (missing 'choices', empty list, missing 'message'/'content') raises an exception whose str(e) becomes the OobaboogaError message with the upstream status code.

Source

Thrown at litellm/llms/oobabooga/chat/transformation.py:66

            original_response=raw_response.text,
            additional_args={"complete_input_dict": request_data},
        )

        ## RESPONSE OBJECT
        try:
            completion_response: Final = raw_response.json()
        except Exception:
            raise OobaboogaError(message=raw_response.text, status_code=raw_response.status_code)
        if "error" in completion_response:
            raise OobaboogaError(
                message=completion_response["error"],
                status_code=raw_response.status_code,
            )
        else:
            try:
                model_response.choices[0].message.content = completion_response["choices"][0]["message"]["content"]
            except Exception as e:
                raise OobaboogaError(
                    message=str(e),
                    status_code=raw_response.status_code,
                )

        model_response.created = int(time.time())
        model_response.model = model
        usage: Final = Usage(
            prompt_tokens=completion_response["usage"]["prompt_tokens"],
            completion_tokens=completion_response["usage"]["completion_tokens"],
            total_tokens=completion_response["usage"]["total_tokens"],
        )
        setattr(model_response, "usage", usage)
        return model_response

    def validate_environment(
        self,
        headers: dict,
        model: str,

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Inspect the exception message (it shows the KeyError/TypeError path) and the logged original_response to see the actual JSON shape.
2. Update text-generation-webui so its OpenAI extension returns proper chat.completion JSON, or pin LiteLLM to a version matching your webui version.
3. Ensure api_base points at /v1-compatible routes and a chat-capable model is loaded.

try:
    resp = litellm.completion(model="oobabooga/m", messages=msgs, api_base=base)
except Exception as e:
    if "choices" in str(e) or "message" in str(e):
        # response shape mismatch: dump server response for diagnosis
        dump_raw_response(base)

When it happens

Trigger: The webui returns JSON without a choices array (e.g. a bare error object without an 'error' key, or a legacy response format), or with an empty choices list, or a completion-style response ({'text': ...}) served on the chat endpoint.

Common situations: Version mismatch between text-generation-webui's OpenAI extension and LiteLLM's expected OpenAI chat shape; pointing the chat handler at a legacy/non-chat endpoint; or server stubs returning truncated responses in tests.

Related errors

• No image URL in BFL result
• No polling_url in BFL response
• OCI embed response does not match expected schema: {e}
• raw_response.text
• completion_response["error"]

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/bae12149f9646f11. Report an issue: GitHub.